Research On Blind Equalization Algorithms For High-order QAM Signal

As one of the key technologies in high-order QAM demodulation systems, the blind equalization technique doesn’t need to transmit training sequences periodically. Instead, it can make merely good use of the statistical characteristics of received signals to recover initial signals. Blind equalization has some unique advantages over other technologies in multipoint network and non-cooperative communication systems. However, due to the high density of the high-order QAM signal constellation, the conventional Bussgang algorithms sometimes can’t effectively eliminate the inter-symbol interference(ISI) among QAM signals after multipath propagation. Therefore, it is of great significance to investigate blind equalization algorithms with the features of faster convergence speed and lower steady-state error for QAM signals.In this thesis, three new algorithms based on the analysis of traditional blind equalization algorithms are proposed: MCME-MCMA algorithm, MMCMADD-VSS-LMS algorithm and normalized decorrelation algorithm. Simulation results demonstrate the effectiveness of these new algorithms. The main content of this thesis are summarized as follows:1. The characteristics of wireless channels are analyzed and the fundamental principles of blind equalization algorithms are introduced. Several key indicators about blind equalization algorithms are described, such as switching criteria of dual-mode algorithm and the structure of typical Bussgang algorithms. The principle of fractional spaced blind equalization algorithm is also provided. Moreover, the performances of several conventional Bussgang algorithms are evaluated and compared by means of computer simulations under different scenarios. And at the same time, their advantages and disadvantages are presented theoretically.2. Owing to the poor performance in cold start of the CME-MCMA algorithm and the low convergence speed of MCMA algorithm for high-order QAM signals, a dual-mode CME-MCMA algorithm based on decision direct is proposed. The performance simulations show that for 64 QAM signal and 256 QAM signal in severe multipath channel, the proposed algorithm not only enhances fast convergence speed, but also provides 5 to 7d B improvement in steady-state,compared with MCMA algorithm.3. A novel hybrid blind equalization algorithm, called MMCMA-DD-VSS-LMS algorithm, is proposed through the inspiration of MCMA-DD-LMS algorithm. Simulation results show that the number of iterations can be reduced by 1000 compared with MCMA-DD-LMS algorithm, and its steady-state performance has no variation. By introducing a new VSS function, the convergence speed of the modified MCMA algorithm is improved by 4000 symbols and its steady-state error is with a decrease of 2d B compared with the traditional MCMA algorithm.4. An adaptive normalized decorrelation blind equalization algorithm is put forward. This algorithm combines decorrelation method with blind equalization algorithm. It can diminish the correlation of consecutive signal vectors during equalization by having a rational choice to de-correlate incoming signals on the basis of the correlation between before the decorrelation and after the decorrelation of the equalizer. Simulation results validate that this proposed algorithm can accelerate the convergence of the equalization system evidently without deterioration of the steady-state performance.